Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #

435

36
Lifting Foils

A basic element of a propeller blade is the foil section. Figure . visualizes how foil
sections are embedded in a blade. Foil sections are streamlined shapes designed to
generate a lift force perpendicular to the onflow, hence the name lifting foil. Lifting foils
are also used in rudders, stabilizer fins, and sails. In this chapter we discuss geometry
and basic flow patterns across lifting foils. We also introduce thin foil theory which is
one of the most widely used tools in foil design.

Learning Objectives

j At the end of this chapter students will be able to: j

• explain geometry and flow patterns of lifting foils
• define lift and drag coefficients
• understand the assumptions of thin foil theory

36.1 Foil Geometry and Flow Patterns

The extent of a foil from trailing edge to leading edge is called chord length 𝑐 (Figure .). Foil section
The straight tail–nose line connects trailing and leading edge and serves as the 𝑥-axis of geometry
the foil coordinate system. Here we place the origin in the middle of the chord length 𝑐.
Chord length 𝑐 together with thickness distribution 𝑡(𝑥) and camber distribution 𝑓𝑚 (𝑥)
defines the geometry of a foil section. For the leading edge, a radius 𝑟𝐿𝐸 (nose radius)
may be specified to better define the shape of the hydrodynamically important nose.
The trailing edge of foils is comparatively sharp in most cases. However, thickness at the
trailing edge has to be finite for strength and manufacturing reasons. Superscripts + and
− are used to identify points or properties of upper and lower foil contour, respectively.

Points 𝑓 + on the upper foil side are obtained by moving a distance 𝑡(𝑥) upward in a
direction normal to the camber line. The lower foil side is obtained by moving 𝑡(𝑥)
downward normal to the camber line. The following equations define upper + and
lower − foil contour.
( )
𝑥± = 𝑥 ∓ 𝑡(𝑥) sin 𝜃(𝑥) (.)
( )
±
𝑓 = 𝑓𝑚 (𝑥) ± 𝑡(𝑥) cos 𝜃(𝑥) (.)
Fundamentals of Ship Hydrodynamics: Fluid Mechanics, Ship Resistance and Propulsion, First Edition.
Lothar Birk.
© 2019 John Wiley & Sons Ltd. Published 2019 by John Wiley & Sons Ltd.
Companion website: www.wiley.com/go/birk/hydrodynamics

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

436 36 Lifting Foils

Figure 36.1 Definition of foil geometry

The minus sign in the right-hand side of Equation (.) is listed on top because it
applies to the upper side of the foil. 𝜃(𝑥) measures the angle of the camber distribution
𝑓𝑚 (𝑥) with respect to the tail–nose line.
( )
d𝑓𝑚
𝜃(𝑥) = tan−1 (.)
d𝑥

Typically, the ratio 𝑓0 ∕𝑐 of maximum camber 𝑓0 and chord length 𝑐 is less than 2%.
Consequently, the angle 𝜃 is small, which justifies the simplifications sin(𝜃) ≈ 0 and
j cos(𝜃) ≈ 1. This is equivalent to defining the foil contours by moving up or down from j
the line perpendicular to the tail–nose line rather than normal to the camber line.

𝑥± = 𝑥 (.)
𝑓 ± = 𝑓𝑚 (𝑥) ± 𝑡(𝑥) (.)

The documentation of the foil shape should specify which version of the coordinate
definition is employed.
With the simplified foil contour description (.) and (.), thickness and camber
distribution are equivalent to half the difference and half the sum of upper and lower
contour 𝑦-coordinates respectively.

1( + ) 1( + )
𝑡(𝑥) = 𝑓 − 𝑓− 𝑓𝑚 (𝑥) = 𝑓 + 𝑓− (.)
2 2
Owing to the latter equation, the camber line is also referred to as the mean line.
Flow patterns Figure . shows a typical flow pattern and the associated pressure distribution. The
foil is rotated anticlockwise by the angle of attack 𝛼 with respect to the onflow 𝑣0 . On
the upper (back) side of the foil, streamlines bunch together. The continuity equation
tells us that a narrowing of streamlines indicates an acceleration of the flow. Thus, much
of the upper side of the foil sees flow speeds |𝑣| higher than the speed of the onflow |𝑣0 |.
According to Bernoulli’s equation, increased flow speed causes a reduction in pressure.
This is clearly visible in the negative pressure coefficient 𝐶𝑝+ values for the back of the
foil. As stated before, a negative pressure coefficient does not mean the pressure is
negative; rather, it implies that local pressure 𝑝 is smaller than the reference pressure
𝑝0 . Therefore, the alternative name for the back of the foil is suction side. Especially

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

36.1 Foil Geometry and Flow Patterns 437

Figure 36.2 Flow pattern and pressure distribution for a 2D foil section at angle of attack 𝛼

j j

toward the leading edge, we encounter low pressure and high flow speeds when the foil
is operating at large angles of attack.
The opposite happens on the underside of the foil, the pressure side. Flow is decelerating
to a stop at the stagnation point and is slower than the onflow across almost all of the
face side of the foil. The pressure coefficient 𝐶𝑝− is positive. The pressure difference
between upper and lower side of a foil determines its lift force.

Δ𝐶𝑝 = 𝐶𝑝+ − 𝐶𝑝− (.)

Friction between fluid and foil develops into a thin boundary layer. The resulting wall
shear stress creates a drag force.
Note that in Figure ., the depicted angle of attack 𝛼 is rather high with about Stall
12 degrees. Typical operational angles of foils are much smaller. Increases in angle of
attack will cause the foil to stall. During a stall event, the flow separates already at the
leading edge, as illustrated in Figure .. Instead of a flow with high speed and low
pressure, the foils produce a large wake filled with large eddies. The lift force all but
vanishes and drag becomes the dominant force.
Stall causes a sudden loss of lift and flight control in airplanes. If the pilots fail to
counteract immediately, the plane will crash. The stalling of marine propeller blades
has less severe consequences but will also result in loss of performance and increased
vibrations.

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

438 36 Lifting Foils

Figure 36.3 Foil in stalled flow condition

Figure 36.4 Complete vortex system of the foil section

j j
36.2 Lift and Drag
Kutta- The German mathematician Martin Wilhelm Kutta (*–†) and the Russian sci-
Joukowsky’s lift entist Nikolay Yegorovich Joukowsky (*–†, name is anglicanized from Russian)
theorem
derived independently from each other the Kutta-Joukowsky’s lift theorem.

𝐿 = 𝜌 |𝑣0 | Γ (.)

Note that the section lift 𝐿 has the dimension of force per unit length [N/m]. Kutta
is most famous as one of the inventors of the Runge-Kutta method for solving ordi-
nary differential equations and for the Kutta condition which is necessary to describe
lifting flows with potential flow theory. Joukowsky is one of the founding fathers of
aerodynamics.
Lift may be accurately computed with potential flow theory assuming an inviscid,
incompressible, and irrotational fluid. A nonzero circulation Γ ≠ 0 requires the curl
of the velocity vector to vanish as well, i.e. ∇ × 𝑣 ≠ 0. This seems to contradict the
application of potential flow. However, as explained in Chapter , the condition of
vanishing curl only applies to simply connected regions. If we include the foil, the flow
region becomes multiply connected.
The effect of circulation on a lifting flow can be modeled with vortices. According to
Helmholtz’s theorems, vortex strength must be constant in inviscid fluids. So, where
did the circulation come from if it was not there already (foil at rest)? When the foil
is accelerating and circulation is established around the foil, a start-up vortex is left
behind (Figure .). The start-up vortex is of equal strength but rotates in the opposite

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

36.2 Lift and Drag 439

Figure 36.5 Forces acting on the foil section at angle of attack 𝛼

direction compared with the circulation that is traveling with the foil. Therefore, the
sum of vorticity in the entire flow field is still zero.
Although this is a crude model of the physics, start-up vortices are real. The start-up
vortex left behind by a big commercial jet at takeoff may easily flip over a small single
propeller aircraft that happens to fly through it. It is one of the reasons why aircraft of
j different sizes need longer time periods between landing and takeoff than aircraft of j
equal size. The start-up vortex is dissolved by friction over time, and the vortices are
blown away by the wind.
Figure . shows lift force 𝐿 and drag force 𝐷 acting on a foil section. Lift acts Lift and drag
perpendicular to the onflow. However, an additional drag force 𝐷 exists in viscous force
(real) fluids which acts in the direction of the onflow 𝑣0 . Viscous pressure drag for thin
foils is small as long as the angle of attack is small. It increases with increasing angle of
attack 𝛼. Frictional drag is the major drag component and also depends on the surface
roughness of the foil. Increased surface roughness due to fouling or damage is especially
detrimental to performance if it affects the leading edge and the first quarter of the
suction side. See Section . for details.
Besides lift and drag forces there is also a rotational moment which aspires to flip the Pitch moment
foil over. Aircraft have a tail unit to counterbalance the moment. The moment is of
interest for the design of controllable pitch propellers.
Hydrodynamic properties of foil sections are often provided as lift and drag coefficients Lift and drag
𝐶𝑙 and 𝐶𝑑 respectively. coefficients
𝐿
𝐶𝑙 = (.)
1 2
𝜌 𝑣0 𝑐
2
𝐷
𝐶𝑑 = (.)
1 2
𝜌 𝑣0 𝑐
2
These two-dimensional foil section coefficients use the chord length 𝑐 as reference
length. Note that this is different from the three-dimensional resistance coefficients

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

440 36 Lifting Foils

0.025
smooth surface
rough surface

[]
section drag coefficient Cd 0.020

0.015
Typical range
of design lift
coefficients

0.010

Cd =0.008 for initial design

0.005

0.000
1.5 1.0 0.5 0.0 0.5 1.0 1.5
section lift coefficient Cl []

Figure 36.6 Typical lift–drag curves for a thin, cambered foil

used in ship hydrodynamics which use the wetted surface of the hull as reference area
j rather than a projection onto a plane. j
Lift–drag ratio The quality of a foil section may be expressed as the lift–drag ratio 𝐿∕𝐷 or 𝐶𝑙 ∕𝐶𝑑 . The
more lift is generated per unit drag, the more effective is the foil. Form drag of the foil
section increases with the angle of attack 𝛼 and the lift–drag ratio decreases. This may
be best visualized in a plot of drag coefficient 𝐶𝑑 over lift coefficient 𝐶𝑙 as illustrated in
Figure .. The graph shows typical section drag coefficient curves for a smooth and a
rough foil section. The foil has a small camber and its most efficient range of operation
is for section lift coefficients between 0.0 and 0.5. Lift–drag curves for many standard
foil section shapes can be found in Abbott and von Doenhoff ().
In the following we will discuss a method suitable for analyzing the potential flow
around two-dimensional foil sections. It is surprisingly accurate in predicting the lift
force of foil sections. Drag has to be determined from model tests or by semi-empirical
formulas which combine theory and experimental results. A section drag coefficient of
𝐶𝑑 = 0.008 is often used for initial propeller design.

36.3 Thin Foil Theory

Conformal Conformal mapping is the most accurate method for the analysis of two-dimensional
mapping potential flow around foil sections. Conformal mapping transforms the known solution
of the flow around a cylinder into the flow around a foil. The geometries of cylinder and
foil are represented in complex planes 𝑧 = 𝑥 + i𝑦 and 𝜁 = 𝜉 + i𝜂 respectively. At the
heart of the procedure is a complex transformation function which maps points of the 𝑧-
plane (cylinder) into points of the 𝜁-plane (foil). Finding the appropriate transformation

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

36.3 Thin Foil Theory 441

Figure 36.7 Setup of boundary value problem for a thin foil operating at angle of attack 𝛼

function for a given geometry is the most difficult task. Readers interested in lifting
flow are encouraged to study conformal mapping. Detailed descriptions are found in
Abbott and von Doenhoff () and Katz and Plotkin ().
Although conformal mapping is a powerful tool for the analysis of two-dimensional Thin foil theory
forms, it is less well suited for design purposes. In conformal mapping, hydrodynamic assumptions
properties are evaluated for a given shape. During design of lifting foils we would like
to specify desired dynamic characteristics first, e.g. a section lift coefficient, and then
find the geometry which produces this characteristic. A suitable tool for this purpose is
thin foil theory. We assume that the foil geometry is given in the form of a mean line
j 𝑓𝑚 (𝑥) and a thickness distribution 𝑡(𝑥). Offsets of upper and lower foil contour 𝑓 ± (𝑥) j
may be derived from Equations (.) and (.). A foil is considered thin if its slope
with respect to the 𝑥-axis (tail–nose line) is small.
𝜕𝑓 ±
≪1 for 𝑥 ∈ [−𝑐∕2, 𝑐∕2] (.)
𝜕𝑥
In addition, thin foil theory assumes that the angle of attack 𝛼 is small.
𝛼 ≪ 1 angle of attack in radians (.)

36.3.1 Thin foil boundary value problem

Analogous to the analysis of the flow around a cylinder, we seek the velocity potential Φ
as the solution of a boundary value problem. We use a body fixed < 𝑥, 𝑦 > coordinate
system, with the 𝑥-axis falling onto the tail–nose line and pointing from trailing edge to
leading edge (Figure .). The velocity potential Φ is initially comprised of two parts:

• a uniform parallel flow in the far field which is oriented at the angle of attack 𝛼
with respect to 𝑥-axis. It has the velocity potential
𝜙∞ (𝑥, 𝑦) = −𝑣0 𝑥 cos(𝛼) + 𝑣0 𝑦 sin(𝛼) (.)
The velocity 𝑣0 is the speed of the foil relative to the fluid in the far field.
• a disturbance caused by the foil which fades away in the far field. Its potential
𝜙(𝑥, 𝑦) has yet to be determined.

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

442 36 Lifting Foils

We exploit again the superposition principle grounded in the linearity of the Laplace
equation. The total flow potential is simply the sum of its parts.

Φ(𝑥, 𝑦) = 𝜙∞ (𝑥, 𝑦) + 𝜙(𝑥, 𝑦) (.)

The boundary of the fluid domain 𝑉 consists of the foil contour 𝑆𝑏 , the far field 𝑆∞ ,
and the wake barrier 𝑆𝑤 (Figure .). The latter is needed to determine the circulation
Γ. As before, the normal vector 𝑛 is pointing out of the fluid domain. That means 𝑛 is
pointing into the foil contour on 𝑆𝑏 .
Boundary value The boundary value problem consists of the Laplace equation as the field equation
problem representing conservation of mass of an ideal, irrotational fluid, and the boundary
condition for each boundary.

𝚫Φ = 0 in 𝑉
𝜕Φ
= 0 body boundary condition for 𝑆𝑏 (.)
𝜕𝑛
𝜕𝜙
= 0 far field condition for 𝑆∞
𝜕𝑛
plus a Kutta condition for the wake 𝑆𝑤 . We will specify a Kutta condition in more detail
when we solve the lifting flow problem in Chapter . The body boundary condition
requires that no fluid flows through the foil contour 𝑆𝑏 . This means that the velocity
component normal to the contour has to vanish and the contour itself becomes a
j streamline. The far field condition requires that the disturbance potential 𝜙 vanishes j
far away from the foil.

36.3.2 Thin foil body boundary condition

Physical The body boundary condition holds the key to a significant simplification of the problem.
interpretation Therefore, it is discussed in greater detail. The body boundary condition in the boundary
value problem (.) requests that the normal derivative of the potential vanishes.
The normal derivative is the component of the gradient of the potential that points in
the direction of the normal vector. It is computed via the dot product, and the body
boundary condition requires that this component is of zero length, i.e. the two vectors,
normal and gradient of the potential, are perpendicular to each other.
𝜕Φ
= 𝑛𝑇 ∇ Φ = 0 (.)
𝜕𝑛
Since the gradient of the velocity potential is equal to the velocity vector, the latter must
be tangent to the foil contour. No flow passes through the foil surface.
Normal vector First, we need to relate the normal vector to the geometry of the foil contour. As for the
discussion of the free surface condition in linear wave theory, we make use of implicit
functions. As before, we use superscripts + and − to identify properties of upper and
lower foil contour, respectively.

upper foil surface 𝐹 + (𝑥, 𝑦) = 𝑦 − 𝑓 + (𝑥) = 0 (.)

lower foil surface 𝐹 − (𝑥, 𝑦) = 𝑦 − 𝑓 − (𝑥) = 0 (.)

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

36.3 Thin Foil Theory 443

Figure 36.8 Definition of normal vectors for

upper and lower foil surface

The implicit definition represents an isoline of 𝐹 ± , here 𝐹 ± = const. = 0. The gradient

of a function is pointing in the direction of steepest ascent which is equivalent to being
normal to the isolines of the function. Then, the normal vector of an implicit function
must be parallel to its gradient.

∇𝐹±
𝑛± = ∓ (.)
|∇ 𝐹 ± |

Since the contours 𝑓 ± are only functions of 𝑥, the gradients of 𝐹 + and 𝐹 − are equal to
j j
⎛ 𝜕𝐹 ±
⎞ ±
⎛ − 𝜕𝑓 ⎞
⎜ 𝜕𝑥 ⎟ ⎜ 𝜕𝑥 ⎟
∇𝐹± = ⎜ ± ⎟ = ⎜ (.)
⎜ 𝜕𝐹 ⎟ ⎟
⎝ 𝜕𝑦 ⎠ ⎝ 1 ⎠

As a consequence, the normal vector 𝑛+ on the suction side becomes

+
⎛ 𝜕𝑓 ⎞
1 ⎜ 𝜕𝑥 ⎟
𝑛 +
= √ (.)
( )2 ⎜ ⎟
𝜕𝑓 + ⎝ −1 ⎠
1+
𝜕𝑥

and the normal vector on the pressure side is

−
⎛ − 𝜕𝑓 ⎞
1 ⎜ 𝜕𝑥 ⎟
𝑛− = √ (.)
( )2 ⎜ ⎟
𝜕𝑓 − ⎝ 1 ⎠
1+
𝜕𝑥

Note the different signs to make the normal vector point into the foil (out of the fluid
domain 𝑉 ) on both suction and pressure side (see Figure .).
Substitution of the normal vector (.) into the body boundary condition (.) Suction side
yields:

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

444 36 Lifting Foils

( ) ⎛ 𝜕Φ ⎞
( )
+ 𝑇 𝜕𝑓 + ⎜ 𝜕𝑥 ⎟
, −1 ⎜ (.)
𝜕Φ ⎟
𝑛 ∇Φ = = 0
𝜕𝑥 ⎜ ⎟
⎝ 𝜕𝑦 ⎠
𝜕𝑓 + 𝜕Φ 𝜕Φ
− = 0 on 𝑓 + (.)
𝜕𝑥 𝜕𝑥 𝜕𝑦
In addition, we split the total velocity potential into parallel flow and disturbance
potential
( )
𝜕𝑓 + 𝜕𝜙+ 𝜕𝜙+
−𝑣0 cos(𝛼) + − 𝑣𝑜 sin(𝛼) − = 0 (.)
𝜕𝑥 𝜕𝑥 𝜕𝑦
For a flat plate the change of velocity 𝜕𝜙∕𝜕𝑥 along the plate would vanish. For a thin foil
it will not be zero but significantly smaller than the onflow velocity 𝑣0 . Together with
the basic assumptions (.) and (.) of thin foil theory we introduce the following
simplifications:
𝜕𝑓 + 𝜕𝜙+
≪ 1, cos(𝛼) ≈ 1, sin(𝛼) ≈ 𝛼, and ≪ 𝑣0 (.)
𝜕𝑥 𝜕𝑥
As a consequence, the first parenthesis in Equation (.) reduces to −𝑣0 .
j 𝜕𝑓 + 𝜕𝜙+ j
− 𝑣0 − 𝛼 𝑣𝑜 − = 0 (.)
𝜕𝑥 𝜕𝑦

We rearrange the terms to have the unknown 𝑦-derivative of the disturbance potential
on the left-hand side.
𝜕𝜙+ 𝜕𝑓 +
= − 𝑣 − 𝛼 𝑣𝑜 (.)
𝜕𝑦 𝜕𝑥 0
This is sometimes called the linearized body boundary condition for thin foil theory,
although technically the body boundary condition (.) is already linear. Nonetheless,
the body boundary condition in its form (.) for the upper foil surface represents a
significant simplification. All terms on the right-hand side are known.
Pressure side Repeating the conversion of Equations (.) through (.) for the pressure side
𝑓 − results in the same simplified body boundary condition except that 𝑓 + is replaced
by 𝑓 − .
𝜕𝜙− 𝜕𝑓 −
= − 𝑣 − 𝛼 𝑣𝑜 (.)
𝜕𝑦 𝜕𝑥 0

or in short the body boundary condition for a thin foil now reads as

𝜕𝜙± 𝜕𝑓 ±
= − 𝑣 − 𝛼 𝑣𝑜 for 𝑦 = ±0, 𝑥 ∈ [−𝑐∕2, 𝑐∕2] (.)
𝜕𝑦 𝜕𝑥 0
We satisfy this condition on the upper and lower sides of the 𝑥-axis with 𝑦 = ±0 because
the right-hand side is a univariate function of 𝑥.

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

36.3 Thin Foil Theory 445

Introducing the definition (.) of foil geometry into the thin foil body boundary Thin foil body
condition (.) allows us to split the right-hand side into three distinct parts. With boundary
condition
𝑓 ± = 𝑓𝑚 + 𝑡 we get:
( )
𝜕𝜙± 𝜕𝑓𝑚 𝜕𝑡
= − ± 𝑣0 − 𝛼 𝑣𝑜
𝜕𝑦 𝜕𝑥 𝜕𝑥
𝜕𝑡 𝜕𝑓𝑚
= ∓ 𝑣0 − 𝑣 − 𝛼 𝑣0 for 𝑦 = ±0 (.)
𝜕𝑥 𝜕𝑥 0
This condition has to be satisfied on the tail–nose line between trailing and leading
edge, i.e. 𝑥 ∈ [−𝑐∕2, 𝑐∕2].

36.3.3 Decomposition of disturbance potential

Each term on the right-hand side of Equation (.) depends on a different foil charac- Decomposition
teristic. of disturbance
potential

(i) The first term is a function of the slope of the thickness distribution 𝜕𝑡(𝑥)∕𝜕𝑥.
(ii) The second term is defined by the slope of the camber distribution 𝜕𝑓𝑚 (𝑥)∕𝜕𝑥.
(iii) The last term depends on the angle of attack 𝛼.

j The superposition principle for solutions of the Laplace equation enables us to subdivide j
the disturbance potential into three parts.

𝜙 = 𝜙1 + 𝜙2 + 𝜙3 (.)

The original boundary value problem is split into three separate boundary value prob-
lems. Table . summarizes their key points. In practice, the two lifting flow problems
are solved together. The potential 𝜙1 depends only on the thickness distribution 𝑡(𝑥) Displacement
and describes the flow around a symmetric foil. This is called the displacement flow flow
problem or thickness problem. For zero angle of attack the flow will be symmetric as
well and we do not expect a resultant force acting normal to the flow. The potentials 𝜙2 Lifting flow
and 𝜙3 capture the flow effects caused by camber and the angle of attack, respectively.
In both cases, the flow will be asymmetric to the 𝑥-axis and a lift force will be generated.
Together, 𝜙2 and 𝜙3 describe the lifting flow. In the following two chapters we will solve
displacement and lifting flow problems under the condition that the foil is thin. We
will start with the displacement flow. For the lifting flow we also have to address the
question of the circulation strength Γ.

References
Abbott, I. and von Doenhoff, A. (). Theory of wing sections – including a summary
of airfoil data. Dover Publications, Inc., New York, NY, Dover edition.
Katz, J. and Plotkin, A. (). Low-speed aerodynamics. Cambridge Aerospace Series.
Cambridge University Press, New York, NY, second edition.

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

446 36 Lifting Foils

Table 36.1 The three subtasks of thin foil theory

displacement flow lifting flow

Symmetric foil at 𝛼 = 0 Cambered plate at 𝛼 = 0 Flat plate at angle of attack 𝛼
𝚫𝜙1 = 0 𝚫𝜙2 = 0 𝚫𝜙3 = 0

𝜕𝜙±
1 𝜕𝑡 𝜕𝜙±
2 𝜕𝑓 𝜕𝜙±
3
= ∓ 𝑣0 on 𝑦 = ±0 = − 𝑚 𝑣0 on 𝑦 = ±0 = −𝛼 𝑣0 on 𝑦 = ±0
𝜕𝑦 𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦

∇ 𝜙1 ⟶ 0 for 𝑥 → ±∞ ∇ 𝜙2 ⟶ 0 for 𝑥 → ±∞ ∇ 𝜙3 ⟶ 0 for 𝑥 → ±∞

Kutta condition for Γ Kutta condition for Γ

Neither camber nor angle of at- The BVP for 𝜙2 involves neither The geometry of the foil (𝑓𝑚 (𝑥)
tack is present in this problem. It thickness distribution nor the an- and 𝑡(𝑥)) does not appear in this
describes the flow around a sym- gle of attack. Only the mean line problem. The potential 𝜙3 repre-
metric foil with thickness distri- 𝑓𝑚 (𝑥) is required. The problem sents the flow over a flat plate at
bution 𝑡(𝑥). defines the flow over a cambered angle of attack 𝛼.
plate at zero angle of attack.

j j

Self Study Problems

. Explain stall in your own words. When does it occur and what are its effects?
. Summarize the underlying assumptions of thin foil theory.
. The body boundary condition for the pressure side of a foil requires that the
velocity normal vanishes.
( )
𝜕𝑓 − 𝜕𝜙− 𝜕𝜙−
− −𝑣0 cos(𝛼) + + 𝑣𝑜 sin(𝛼) + = 0
𝜕𝑥 𝜕𝑥 𝜕𝑦

Apply the assumptions of thin foil theory to derive the body boundary condition
for thin foil theory (.).
. Research the geometry of the NACA -digit series and the NACA -xxx series
of foils. Learn how to extract important foil characteristics from the specification,
i.e. which specifications are embedded in the designation NACA . Use the
provide equations to plot the shape of NACA .

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

447

37
Thin Foil Theory – Displacement Flow

In this chapter we continue to discuss the potential flow around a thin foil. The upper
and lower sides of the foil are represented by a mean line 𝑓𝑚 and a symmetric thickness
distribution.
𝑓 ± (𝑥) = 𝑓𝑚 (𝑥) ± 𝑡(𝑥) for 𝑥 ∈ [−𝑐∕2, 𝑐∕2] (.)
By introducing the requirements that

• the angle of attack 𝛼 is small,

𝜕𝑓 ±
• the foil represented by 𝑓 ± (𝑥) is thin with ≪ 1, and
𝜕𝑥
j • any velocities induced by the disturbance potential 𝜙 are small compared to the j
| 𝜕𝜙 |
parallel stream velocity || || ≪ 𝑣0
| 𝜕𝑥 |

we have been able to split the flow problem into three separate boundary value problems.
Here, we solve the first of these boundary value problems which describes the flow
around a symmetric foil with finite but small thickness–chord length ratio 𝑡∕𝑐, no
camber, and at zero angle of attack.

Learning Objectives
At the end of this chapter students will be able to:

• state and solve boundary value problems

• understand and apply singularity distributions

• discuss and compute the pressure distributions around symmetric foils

37.1 Boundary Value Problem

The flow around the symmetric foil at zero angle of attack may be represented by a
velocity potential Φ which consists of two parts: A flow potential 𝜙∞ , which describes

Fundamentals of Ship Hydrodynamics: Fluid Mechanics, Ship Resistance and Propulsion, First Edition.
Lothar Birk.
© 2019 John Wiley & Sons Ltd. Published 2019 by John Wiley & Sons Ltd.
Companion website: www.wiley.com/go/birk/hydrodynamics

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

448 37 Thin Foil Theory – Displacement Flow

Figure 37.1 The boundary value problem of a symmetric thin foil with finite thickness and zero
angle of attack

the flow in the far field, and a velocity potential 𝜙1 , which describes the disturbance
caused by the foil. The potentials are functions of the planar coordinates 𝑥 and 𝑦.
Φ(𝑥, 𝑦) = 𝜙∞ (𝑥, 𝑦) + 𝜙1 (𝑥, 𝑦) (.)

Figure . shows the control volume and its boundaries. The boundary value problem
j for the disturbance potential consists of the Laplace equation and appropriate boundary j
conditions.
𝚫𝜙1 (𝑥, 𝑦) = 0 for (𝑥, 𝑦) ∈ 𝑉
𝜕𝜙1 [ ]
d𝑡 𝑐 𝑐
= ∓𝑣0 for 𝑥 ∈ − , and 𝑦 = 0± (.)
𝜕𝑦 d𝑥 2 2
√
𝜙1 (𝑥, 𝑦) ⟶ const. for 𝑥2 + 𝑦2 → ∞
Note that we simplified the body boundary condition according to thin foil theory
assumptions.
This problem is similar to the flow around a cylinder or a Rankine oval. The flow around
the latter was described by the potentials of a source, a sink, and parallel. This lends
credit to the idea that more complicated shapes may be represented by a multitude of
sources and sinks. As long as the sum of their strengths vanishes we will obtain a closed
contour.
Line distribution Going one step further, we use a line of distributed sources instead of a set of discrete
of sources sources and sinks. This is the same concept as a distributed load in mechanics of
materials. The source strength 𝜎 is now given as a ‘volume’ flow per unit length. Note
that the ‘volume’ flow in the plane is measured in length units squared per second.
We place the sources on the chord line 𝑦 = 0 between 𝑥 = −𝑐∕2 and 𝑥 = +𝑐∕2. The
total disturbance potential is obtained by integration along the length of the source
distribution.
+𝑐∕2
𝜎(𝜉) (√ )
𝜙1 (𝑥, 𝑦) = ln (𝑥 − 𝜉)2 + 𝑦2 d𝜉 (.)
∫ 2𝜋
−𝑐∕2

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

37.1 Boundary Value Problem 449

A pair of coordinates (𝑥, 𝑦) marks the point where we want to know the potential or
velocity. The integration variable 𝜉 marks the position of sources along the tail–nose
line (𝜂 = 0). We have to adjust the yet unknown source strength 𝜎(𝜉) so that the closed
contour looks like our thin foil and represents a streamline of the flow. Therefore we
will employ the linearized body boundary condition for the displacement flow from
Table . to find the source strength.
The derivative of the disturbance potential (.) with respect to 𝑦 is part of the body Transverse
boundary condition. In the potential, both the limits of the integral and the integra- velocity
tion variable are independent of 𝑦, which allows us to interchange integration and
differentiation.
+𝑐∕2
𝜕𝜙1 (𝑥, 𝑦) 𝜎(𝜉) 𝑦
= ( ) d𝜉 (.)
𝜕𝑦 ∫ 2𝜋 (𝑥 − 𝜉)2 + 𝑦2
−𝑐∕2

The linearized body boundary condition is satisfied over the tail–nose line on the 𝑥- Evaluating the
axis with 𝑦 = 0 instead of the actual foil contour. The integrand in (.) vanishes integral
everywhere for 𝑦 = 0 except for the position 𝑥 = 𝜉 for which we obtain an undefined
expression of the type ‘/.’ Without loss of accuracy, we may limit the extent of the
integral to some finite interval of length 2𝛿 around 𝑥. 𝛿 is a small positive number.
The regions outside of this area will make no contribution to the integral because the
integrand vanishes (𝑦 = 0).
𝑥+𝛿
j 𝜕𝜙1 (𝑥, 𝑦) 𝜎(𝜉) 𝑦 j
= ( ) d𝜉 (.)
𝜕𝑦 ∫ 2𝜋 (𝑥 − 𝜉)2 + 𝑦2
𝑥−𝛿

We make 𝛿 so small that it is safe to assume that the source strength takes the value at
𝑥, i.e. 𝜎(𝜉) = 𝜎(𝑥) for 𝜉 ∈ [𝑥 − 𝛿, 𝑥 + 𝛿]. The source strength becomes independent of
the integration variable and can be brought in front of the integral sign.
𝑥+𝛿
𝜕𝜙1 (𝑥, 𝑦) 𝜎(𝑥) 𝑦
= ( ) d𝜉 (.)
𝜕𝑦 2𝜋 ∫ (𝑥 − 𝜉)2 + 𝑦2
𝑥−𝛿

We will proceed with caution and search for the value of integral (.) by letting
𝑦 = 0 ± 𝜀 and compute the limits for 𝜀 → +0 and 𝜀 → −0 if they exist.
𝑥+𝛿
𝜕𝜙1 (𝑥, 𝑦 = ±0) 𝜎(𝑥) 𝜀
= lim ( ) d𝜉 (.)
𝜕𝑦 𝜀→±0 2𝜋 ∫ (𝑥 − 𝜉)2 + 𝜀2
𝑥−𝛿

The following variable substitution simplifies the integration:

d𝜆
𝜆 = (𝑥 − 𝜉) = −1 i.e. d𝜉 = −d𝜆
d𝜉

The limits of the integral transform into

𝜆𝑙 = 𝑥 − (𝑥 − 𝛿) = +𝛿 𝜆𝑢 = 𝑥 − (𝑥 + 𝛿) = −𝛿

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

450 37 Thin Foil Theory – Displacement Flow

Figure 37.2 The inverse tangent function

2

arctan()
0

2
60 40 20 0 20 40 60

Substituting 𝜆 for 𝜉 changes the integral changes into the following form:

−𝛿
𝜕𝜙1 (𝑥, 𝑦 = ±0) 𝜎(𝑥) 𝜀
= lim ( ) (−d𝜆) (.)
𝜕𝑦 𝜀→±0 2𝜋 ∫ 𝜆2 + 𝜀2
+𝛿

j We switch upper and lower limits and consequently remove the minus sign. j
+𝛿
𝜕𝜙1 (𝑥, 𝑦 = ±0) 𝜎(𝑥) 𝜀
= lim ( ) d𝜆 (.)
𝜕𝑦 2𝜋 𝜀→±0 ∫ 𝜆2 + 𝜀2
−𝛿

The antiderivative of the integral of type (.) is the inverse tangent function arctan()
(Figure .).

𝜕𝜙1 (𝑥, 𝑦 = ±0) 𝜎(𝑥) [ ( 𝜂 )]+𝛿

1
= lim 𝜀 arctan
𝜕𝑦 2𝜋 𝜀→±0 𝜀 𝜀 −𝛿
𝜎(𝑥) ( ( ) ( ))
𝛿 −𝛿
= lim arctan − arctan
2𝜋 𝜀→±0 𝜀 𝜀

If 𝜀 approaches zero from the positive side, the first arctan() function’s argument will
grow toward +∞. Then the function value becomes +𝜋∕2 (see Figure .). The
argument of the second arctan() grows toward −∞ and the tangent value approaches
−𝜋∕2.

𝜕𝜙1 (𝑥, 𝑦 = +0) 𝜎(𝑥) ( ( ) ( ))

𝛿 −𝛿
= lim arctan − arctan
𝜕𝑦 2𝜋 𝜀→+0 𝜀 𝜀
𝜎(𝑥) ( 𝜋 −𝜋 )
= −
2𝜋 2 2
𝜎(𝑥)
= + (.)
2

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

37.1 Boundary Value Problem 451

For the case where 𝜀 approaches zero from the negative side, the signs reverse for the
arguments of the arctan() functions.
𝜕𝜙1 (𝑥, 𝑦 = −0) 𝜎(𝑥) ( ( ) ( ))
𝛿 −𝛿
= lim arctan − arctan
𝜕𝑦 2𝜋 𝜀→−0 𝜀 𝜀
𝜎(𝑥) ( −𝜋 𝜋 )
= −
2𝜋 2 2
𝜎(𝑥)
= − (.)
2
In summary:
𝜕𝜙1 (𝑥, +0) 𝜎(𝑥) 𝜕𝜙1 (𝑥, −0) 𝜎(𝑥)
𝑣+
1
= = + and 𝑣−
1
= = − (.)
𝜕𝑦 2 𝜕𝑦 2

This result needs some physical interpretation: Normal velocity

induced by line
• On the upper side of the foil with 𝑦 = +0 the source distribution generates an sources
upward oriented transverse disturbance velocity of 𝑣+
1
= +𝜎(𝑥)∕2.
• On the lower side of the foil where 𝑦 = −0 the transverse velocity 𝑣−
1
= −𝜎(𝑥)∕2
is oriented downward with the same magnitude.

Thus the velocity discontinuously changes when crossing from one side of the foil to
the other. The total change in velocity is equal to the source strength 𝜎(𝑥). Although
j this was derived here for a line distribution of sources it is a general feature of source j
distributions.
Finally, we introduce the result (.) into the linearized body boundary condition (.). Exploiting the
body boundary
𝜕𝜙1 d𝑡 condition
= ∓𝑣0 (.)
𝜕𝑦 d𝑥
Therefore, the source strength is proportional to the slope of the thickness distribution:
d𝑡(𝑥)
𝜎(𝑥) = −2𝑣0 (.)
d𝑥
Both the upper and lower sides yield the same result which is expected since we have
only one source distribution.
Figure . shows a symmetric foil section and the distribution of source strength over Source strength
the chord of the foil. The source strength will be negative toward the trailing edge; the distribution
sources are actually sinks. At the position of maximum thickness with d𝑡∕d𝑥 = 0 the
source strength is zero and then grows positive toward the leading edge. If the nose is
round and meets the chord line (𝑦 = 0) at a right angle the source strength distribution
will have a singularity. Clearly, |d𝑡∕d𝑥| → ∞ is in violation of our thin foil assumption.
A correction is needed to represent the flow correctly in the vicinity of the leading edge
(see e.g. Lighthill, ).
With the source strength distribution known, we can restate the disturbance potential.
+𝑐∕2
𝑣 d𝑡(𝜉) (√ )
𝜙1 (𝑥, 𝑦) = − 0 ln (𝑥 − 𝜉)2 + 𝑦2 d𝜉 (.)
𝜋 ∫ d𝜉
−𝑐∕2

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

452 37 Thin Foil Theory – Displacement Flow

Figure 37.3 The source strength distribution 𝜎(𝜉) as a function of the slope d𝑡∕d𝑥 of the foil
surface

Since the effect of sources declines with growing distance, the disturbance potential
will satisfy the far field condition of the boundary value problem (.).
Total potential Together with the parallel flow we have a velocity potential for the flow around our thin,
for thickness symmetric foil at zero angle of attack.
problem
j +𝑐∕2 j
𝑣 d𝑡(𝜉) (√ )
Φ(𝑥, 𝑦) = −𝑣0 𝑥 − 0 ln (𝑥 − 𝜉)2 + 𝑦2 d𝜉 (.)
𝜋 ∫ d𝜉
−𝑐∕2

37.2 Pressure Distribution

Bernoulli With the potential (.) the velocity distribution in the domain around the foil is
equation for thin known. The Bernoulli equation for steady flow allows us to compute the pressure
foils
distribution. We ignore the gravity forces assuming that foil and fluid are in equilibrium
when at rest.
1 1
𝜌|∇ Φ|2 + 𝑝 = 𝜌 𝑣20 + 𝑝0 (.)
2 2
The magnitude of the velocity squared |∇ Φ|2 is given by
( )2 ( )2
2 𝜕Φ 𝜕Φ
|∇ Φ| = +
𝜕𝑥 𝜕𝑦
( )2 ( )2
𝜕𝜙 𝜕𝜙1 𝜕𝜙1
= 𝑣20 − 2𝑣0 1 + +
𝜕𝑥 𝜕𝑥 𝜕𝑦
The latter two terms must be very small if the foil is thin because they are a function of
(d𝑡∕d𝑥)2 . These small, quadratic terms will be ignored from hereon forward.
𝜕𝜙1
|∇ Φ|2 = 𝑣20 − 2𝑣0 (.)
𝜕𝑥

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

37.2 Pressure Distribution 453

Substitution of this result into (.) yields

𝜕𝜙1
𝑝 − 𝑝0 = 𝜌 𝑣0 (.)
𝜕𝑥
The pressure coefficient 𝐶𝑝 (𝑥) will be
𝑝 − 𝑝0 2 𝜕𝜙1
𝐶𝑝 (𝑥) = = (.)
1 𝑣0 𝜕𝑥
2
𝜌 𝑣20

The 𝑥-derivative of the disturbance potential (.) is

+𝑐∕2
𝜕𝜙1 𝑣 d𝑡(𝜉) (𝑥 − 𝜉)
= − 0 ( ) d𝜉 (.)
𝜕𝑥 𝜋 ∫ d𝜉 (𝑥 − 𝜉)2 + 𝑦2
−𝑐∕2

The pressure coefficient on the foil surfaces, approximated by 𝑦 = ±0, has to be evalu-
ated carefully because the integrand has a singularity for 𝑥 = 𝜉.
+𝑐∕2
2 d𝑡(𝜉) 1
𝐶𝑝 (𝑥) = − − d𝜉 (.)
𝜋 ∫ d𝜉 (𝑥 − 𝜉)
−𝑐∕2

This is called a Cauchy principal value integral, and indicated by the dash through the
middle of the integral sign.
j In this case, it can be shown that the integral exists. We abbreviate d𝑡∕d𝜉 = 𝑡′ and split Cauchy principal j
the integral into three parts. value integral

+𝑐∕2

− 𝑡′ (𝜉) 1 d𝜉
∫ (𝑥 − 𝜉)
−𝑐∕2
𝑥−𝜀 𝑥+𝜀 𝑐∕2
1 1 1
= 𝑡′ (𝜉) d𝜉 + − 𝑡′ (𝜉) d𝜉 + 𝑡′ (𝜉) d𝜉 (.)
∫ (𝑥 − 𝜉) ∫ (𝑥 − 𝜉) ∫ (𝑥 − 𝜉)
−𝑐∕2 𝑥−𝜀 𝑥+𝜀

The first and last integral on the right-hand side do not contain singularities. The case
𝑥 = 𝜉 is not possible. Therefore, the two integrals will exist if 𝑡′ = d𝑡∕d𝜉 is bounded.
The middle integral in Equation (.) needs further investigation. We assume that
𝜀 is so small that we can extract the derivative of the thickness distribution 𝑡′ (𝜉) from
the integral. It will take the value 𝑡′ (𝑥) at position 𝑥 along the tail–nose line. The
remaining integral is a Cauchy principal value integral, and it has a value, if the limit on
the right-hand side exists.
𝑥+𝜀
⎛ 𝑥−𝛿 𝑥+𝜀
⎞
1 ⎜ 1 1
− d𝜉 = lim d𝜉 + d𝜉 ⎟
∫ (𝑥 − 𝜉) 𝛿→+0 ⎜ ∫ (𝑥 − 𝜉) ∫ (𝑥 − 𝜉) ⎟
𝑥−𝜀 ⎝𝑥−𝜀 𝑥+𝛿 ⎠
⎛ 𝑥−𝛿 𝑥+𝜀
⎞
1 1
= lim ⎜ d𝜉 − d𝜉 ⎟ (.)
𝛿→+0 ⎜ ∫ (𝑥 − 𝜉) ∫ (𝜉 − 𝑥) ⎟
⎝𝑥−𝜀 𝑥+𝛿 ⎠

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

454 37 Thin Foil Theory – Displacement Flow

During this conversion the operands switched position in the denominator of the
second integral. The resulting minus sign has been extracted from the integral. This
avoids negative arguments in the integrand, since 𝜉 > 𝑥 for the second integral. This is
helpful in the subsequent integration because the natural logarithm is undefined for
negative arguments.
The remaining integrals can now be solved.
𝑥+𝜀 ([ ]𝑥−𝛿 [ ]𝑥+𝜀 )
− 1
d𝜉 = lim − ln(𝑥 − 𝜉) − ln(𝜉 − 𝑥)
∫ (𝑥 − 𝜉) 𝛿→+0 𝑥−𝜀 𝑥+𝛿
𝑥−𝜀

The additional minus sign in the first antiderivative results from the inner derivative.
Finally, we substitute the limits for the integration variable 𝜉:
𝑥+𝜀
([ ] [ ])
1
− d𝜉 = lim − ln(𝛿) + ln(𝜀) − ln(𝜀) − ln(𝛿) (.)
∫ (𝑥 − 𝜉) 𝛿→+0
𝑥−𝜀
= lim (− ln(𝛿) + ln(𝛿) + ln(𝜀) − ln(𝜀))
𝛿→+0
( )
= lim 0
𝛿→+0
= 0

j The limit for 𝛿 → 0 obviously exists. The value of the integral is zero, i.e. the middle j
integral in Equation (.) makes no contribution to the pressure coefficient. Thus we
have proven that the Cauchy principal value integral for the pressure coefficient (.)
exists, if the derivative of the thickness distribution is well behaved.

37.3 Elliptical Thickness Distribution

Example: As an example, we investigate the elliptical thickness distribution.
elliptical foil
√
section ( )2 [ ]
𝑡max 2𝜉 𝑐 𝑐
𝑡(𝜉) = 1− for 𝜉 ∈ − , (.)
2 𝑐 2 2

Elliptical thickness distributions are not often used, although they are less prone to
cavitation than other profile shapes.
Variable Before substituting the elliptical thickness distribution (.) into the integral for the
substitution pressure coefficient (.), we initiate a variable substitution. The integration variable
𝜉 will be replaced by a polar coordinate 𝜑 such that

𝑐 d𝜉 𝑐 𝑐
𝜉 = cos(𝜑) = − sin(𝜑) d𝜉 = − sin(𝜑) d𝜑 (.)
2 d𝜑 2 2
The integration limits become:
𝑐 𝑐
𝜉𝑙 = − ⟶ 𝜑𝑙 = 𝜋 𝜉𝑢 = ⟶ 𝜑𝑢 = 0 (.)
2 2

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

37.3 Elliptical Thickness Distribution 455

The point (𝑥, 0), where we evaluate the pressure coefficient along the foil, is transformed
in the same way, but it has its own angle 𝜃, independent of the integration variable.
𝑐
𝑥 = cos(𝜃) (.)
2
with 𝜃 ∈ [0, 𝜋].
Over the 𝜑-axis, the elliptical thickness distribution (.) is represented by a sine
curve.
𝑡 √ 𝑡
𝑡(𝜑) = max 1 − cos2 (𝜑) = max sin(𝜑) (.)
2 2

We also need the derivative of the thickness distribution with respect to the integration Derivative of
variable 𝜉. thickness
distribution
d𝑡 d𝑡 d𝜑 𝑡 1
= = max cos(𝜑) ( )
d𝜉 d𝜑 d𝜉 2 d𝜉
d𝜑

and with the derivative d𝜉∕d𝜑 from (.)

d𝑡 𝑡 cos(𝜑)
= max ( )
d𝜉 2 𝑐
− sin(𝜑)
2
j 𝑡 cos(𝜑) j
= − max (.)
𝑐 sin(𝜑)

We introduce the variable substitution (.), the transformed field point coordinate Pressure
(.), and the derivative of the thickness distribution (.) into Equation (.) coefficient
for the pressure coefficient.

0
𝑡 cos(𝜑) ( )
2 1 𝑐
𝐶𝑝 (𝜃) = − − − max [ ] − sin(𝜑) d𝜑 (.)
𝜋∫ 𝑐 sin(𝜑) 𝑐 cos(𝜃) − 𝑐 cos(𝜑) 2
𝜋 2 2
We carefully consolidate the three minus signs into one and extract all constants from
the integral.
0
2𝑡 cos(𝜑) sin(𝜑)
𝐶𝑝 (𝜃) = − max − [ ] d𝜑 (.)
𝜋 𝑐 ∫ sin(𝜑) cos(𝜃) − cos(𝜑)
𝜋

Next, we eliminate the sin(𝜑) in numerator and denominator and extract a factor −1
from the denominator which we convert into a reversal of the integration limits.
𝜋
2𝑡 cos(𝜑)
𝐶𝑝 (𝜃) = − max − [ ] d𝜑 (.)
𝜋 𝑐 ∫ cos(𝜑) − cos(𝜃)
0

The Cauchy principal value integral in equation (.) is not quite so easy to solve.

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

456 37 Thin Foil Theory – Displacement Flow

Glauert integrals Luckily for us, Glauert (, pp. –) solved this as part of a whole group of
integrals, which are known in hydrodynamics as Glauert integrals.
𝜋
cos(𝑛𝜑) sin(𝑛𝜃)
𝐼𝑛 (𝜃) = − [ ] d𝜑 = 𝜋 (.)
∫ cos(𝜑) − cos(𝜃) sin(𝜃)
0

In our case 𝑛 = 1, and the value of the integral 𝐼1 = 𝜋. For the pressure coefficient of
an elliptical thickness distribution we obtain a constant negative value:
2 𝑡max
𝐶𝑝 (𝜃) = − (.)
𝑐
This is somewhat surprising because we expect to see the value 𝐶𝑝 = 1 at the aft
and forward stagnation points. But then, close to the stagnation points our foil shape
violates the primary condition of thin foil theory. At the rounded ends of the elliptical
foil section, the derivative 𝜕𝑓
𝜕𝑥
𝜕𝑡
= 𝜕𝑥 of the shape is anything but small. In fact, it does
not exist and its values approach ±∞.
Examples Overall, the pressure distribution matches the exact potential theory result obtained by
conformal mapping quite well. Figure . shows the results for an elliptical foil with
% thickness to chord length ratio. The minimum 𝐶𝑝 value is slightly under predicted
by thin foil theory, but it is close to its conformal mapping value. As expected, thin foil
theory fails to predict the correct pressure toward trailing and leading edges of the foil.

j One could argue that a foil with 𝑡max ∕𝑐 = 0.10 is not really a thin foil. Figure . shows j
the result for an elliptical foil section with a thickness to chord length ratio of just %.
The minimum 𝐶𝑝 value is fairly accurate and the deviations are much smaller toward
leading and trailing edges.
Figures . and . also show the velocity magnitude as derived by conformal map-
ping. If one plots the magnitude of the velocity derived from thin foil theory, huge
discrepancies are found toward the ends of the foil if the transverse velocity component
𝑣1 = 𝜕𝜙1 ∕𝜕𝑦 is included. As listed in Equation (.), 𝑣1 is a function of the slope
of the thickness distribution. At the ends of the foil, the slope is ±∞ which renders
the velocity prediction as incorrect. Hence, it is common practice to compute the
magnitude of the velocity as in Equation (.) without the parts 𝑢21 and 𝑣21 .
√
2 𝜕𝜙1
|∇ Φ| = 𝑣0 1 − (.)
𝑣0 𝜕𝑥

References
Glauert, H. (). The elements of airofoil and airscrew theory. Cambridge University
Press, Cambridge, United Kingdom, second edition.
Lighthill, M. (). A new approach to thin foil theory. The Aeronautical Quarterly,
():–.
 Hermann Glauert (* – †), British aerodynamicist

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

37.3 Elliptical Thickness Distribution 457

j j
Figure 37.4 Comparison of thin foil theory and conformal mapping (exact) pressure coefficient
for an elliptical foil with thickness to chord length ratio of 𝑡max ∕𝑐 = 0.10

Self Study Problems

. Summarize the equations that define the linearized boundary value problem for
the displacement flow. State the physical meaning of each equation.

. Given is a line distribution of sources along the tail–nose line with strength 𝜎(𝑥).
By how much changes the transverse velocity 𝑣 as you move from the underside
(𝑦 = 0− ) to the upper side of the source distribution (𝑦 = 0+ )?

. Explain in your own words why the source strength at the leading edge is un-
bounded in Figure ..

. Compute and plot the pressure coefficient over the chord length of an elliptical
foil with 7% thickness–chord length ratio. Use the thin foil approximation.

. Compute the value of the following integral for 𝜃 = 𝜋∕3:

𝜋
cos(3𝜑)
𝐼3 (𝜃) = − [ ] d𝜑 (.)
∫ cos(𝜑) − cos(𝜃)
0

j
 Trim Size: mm × mm Single Column Tight Birk — “fshy” — // — : — page  — #
j

458 37 Thin Foil Theory – Displacement Flow

j j
Figure 37.5 Comparison of thin foil theory and conformal mapping (exact) pressure coefficient
for an elliptical foil with thickness to chord length ratio of 𝑡max ∕𝑐 = 0.05

. Derive the velocity distribution induced on a thin foil (𝑥 ∈ [−𝑐∕2, 𝑐∕2]) by a

parabolic thickness distribution
[ ( )2 ]
𝑡 2𝑥 𝑐 𝑐
𝑡(𝑥) = max 1 − for − ≤ 𝑥 ≤
2 𝑐 2 2

The parabolic thickness distribution is used to approximate the flow around ogival
foil sections. Compare the velocity and pressure distribution with the elliptical
thickness distribution.